RAM usage / preload

You're new to the LinuxSampler world? You don't know where to start and nothing works? Here's the place to ask for help.
Post Reply
typewriter
Advanced User
Posts: 147
Joined: Fri Sep 26, 2008 9:04 am

RAM usage / preload

Post by typewriter » Sun Aug 09, 2009 6:31 pm

A small amount of a sample is cached into RAM before streaming a patch from disc, right? How much is precached that way? 64kb? More? Is it varying depending on the gig instrument? With linuxsampler VST for 64bit soon to be released I was wondering how many instruments we could load on a 8 gig Vista 64 PC. I recall a posting in VSL forum that you would need 8 PCs with Gigastudio to load 75.000 samples (which should be about 40% of VSL's First Edition) because there would be a limit of Giga of 1 GB loaded samples per PC (http://community.vsl.co.at/forums/t/19935.aspx). How many LS PCs would be need to achive the same result?

User avatar
cuse
Developer
Posts: 366
Joined: Wed Jan 23, 2008 10:07 pm
Location: Germany

Re: RAM usage / preload

Post by cuse » Sun Aug 09, 2009 7:02 pm

That depends on the samples used. LinuxSampler preloads a fix amount of sample points. This constant value can be defined at compile time by the ./configure script switch --enable-preload-samples=65536 . I think the current Windows binaries use this value (65536). So let's say you have a stereo sample with 16 bit, then the amount of precached data for one sample would be: 2 (stereo) * 2 byte (16 bit) * 65536 = 262144 bytes , but of course only if the sample is large enough.

typewriter
Advanced User
Posts: 147
Joined: Fri Sep 26, 2008 9:04 am

Re: RAM usage / preload

Post by typewriter » Sun Aug 09, 2009 11:49 pm

Okay this means LS can load 4096 16bit stereo samples per gigabyte ram. When I calculated it correctly.

User avatar
cuse
Developer
Posts: 366
Joined: Wed Jan 23, 2008 10:07 pm
Location: Germany

Re: RAM usage / preload

Post by cuse » Mon Aug 10, 2009 6:58 am

Very roughly yes. You still have other overheads which would reduce it, and if you got very small samples, it would increase it.

Post Reply